¹ Throughout this analysis we distinguish between partisan realignments and shifts in the balance of public support between major parties. As Crewe reminds us, the concept of realignment, as defined originally by V. O. Key, refers to more than a landslide victory or a temporary shift in public support from one party to another. Realignment requires a fundamental and enduring shift in partisan loyalty and party identification usually accompanied by a transformation in the social structure of party support and a permanent shift in the balance of power among political interests and ideas. ‘A truly realigning election crystallizes a structural change in the mass basis of political order, much as an earthquake is both cause and effect of a geological shift in the landscape.’ (See Crewe, I., ‘How to Win a Landslide Without Really Trying: Why the Conservatives Won in 1983’, in Ranney, A., ed., Britain at the Polls, 1983 (New York: Duke University Press, 1985), pp. 155–96 at p. 187.)Google Scholar Although a shift in the balance of public support among parties usually accompanies a partisan realignment, it is not a necessary feature of realignment. Neither is a shift in the balance of party support sufficient evidence that realignment has occurred. Although we touch upon realignment issues in this article, our principal concern is with understanding and explaining shifts in the balance of public support between the major parties.

² Representative works include: Butler, D. and Stokes, D., Political Change in Britain (New York: St Martin's Press, 1976), pp. 155–208Google Scholar; Crewe, I., Särlvik, B., and Alt, J., ‘Partisan Dealignment in Britain, 1964–1974’, British Journal of Political Science, 7 (1977), 129–90CrossRefGoogle Scholar; Franklin, M. and Mughan, A., ‘The Decline of Class Voting in Britain: Problems of Analysis and Interpretation’, American Political Science Review, 72 (1978), 523–34CrossRefGoogle Scholar; Särlvik, B. and Crewe, I., Decade of Dealignment (Cambridge: Cambridge University Press, 1983), pp. 30–115Google Scholar; Alt, J., ‘Dealignment and the Dynamics of Partisanship in Britain’, in Dalton, R. J. et al. , eds, Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment (Princeton, NJ: Princeton University Press, 1984), pp. 289–329Google Scholar; Franklin, M., The Decline of Class Voting (Oxford: Oxford University Press, 1985)Google Scholar; and Dunleavy, P. and Husbands, C. T., British Democracy at the Crossroads: Voting and Party Competition in the 1980s (London: Allen & Unwin, 1985).Google Scholar

³ Mishler, W., Hoskin, M. and Fitzgerald, R., ‘“Hunting the Snark:” or Searching for Evidence of that Widely Touted but Highly Elusive Resurgence of Support for Conservative Parties in Britain, Canada, and the United States’, in Cooper, B., Kornberg, A. and Mishler, W., The Resurgence of Conservatism in Anglo-American Democracies (Durham, NC: Duke University Press, 1988), 54–95.Google Scholar

⁴ If, for example, public support for the Conservative party declines by 5 percentage points during a period in which Labour support also declines but by 10 percentage points, the result obviously is a 5 percentage point net ‘swing’ in public support in favour of the Conservatives. The practical effect of such a swing in the short term would be much the same as that produced by a realignment of 2.5 per cent of the electorate from Labour to the Conservative party.

⁵ For a discussion of this literature, see Clarke, H. and Stewart, M., ‘Dealignment of Degree: Partisan Change in Britain, 1974–1983’, Journal of Politics, 46 (1984), 689–718.CrossRefGoogle Scholar

⁶ Miller, W., ‘There Was No Alternative: The British General Election of 1983’, Parliamentary Affairs, 38 (1984), 364–84.CrossRefGoogle Scholar

⁷ See, among numerous others: Crewe, I., ‘Why the Conservatives Won’, in Penniman, H. R., ed., Britain at the Polls, 1979 (Washington, DC: American Enterprise Institute, 1981), pp. 263–305Google Scholar; Butler, D. and Kavanagh, D., The British General Election of 1979 (London: Macmillan, 1980), pp. 315–51CrossRefGoogle Scholar; and Särlvik, and Crewe, , Decade of Dealignment, pp. 247–80.Google Scholar

⁸ Representative citations include: Butler, D. and Kavanagh, D., The British General Election of 1983 (London: Macmillan, 1984), pp. 269–98CrossRefGoogle Scholar; Crewe, , ‘How to Win a Landslide Without Even Trying’, pp. 186–96Google Scholar; Dunleavy, and Husbands, , British Democracy at the Crossroads.Google Scholar

⁹ Crewe, I., ‘What's Left for Labour: An Analysis of Thatcher's Victory’, Public Opinion, 07/08 (1987), 52–6.Google Scholar

¹⁰ Useful summaries of the literature include Paldam, M., ‘A Preliminary Survey of the Theories and Findings of Vote and Popularity Studies’, European Journal of Political Research, 9 (1981), 181–99CrossRefGoogle Scholar, and Clarke, H., Stewart, M. and Zuk, G., eds, Politics, Power, and the Economy: Canada, Great Britain, and the United States (Pittsburgh: Pittsburgh University Press, 1988, forthcoming), Chapter 1.Google Scholar

¹¹ See, in particular, Alt, J., The Politics of Economic Decline (Cambridge: Cambridge University Press, 1979)Google Scholar; Hibbs, D., ‘Economic Outcomes and Political Support for British Governments among Occupational Classes’, American Political Science Review, 76 (1982), 259–79CrossRefGoogle Scholar; Whiteley, P., ‘Inflation, Unemployment and Government Popularity: A Dynamic Model for the United States, Britain and West Germany’, Electoral Studies, 3 (1984), 25–46CrossRefGoogle Scholar; Whiteley, P., ‘Macroeconomic Performance and Government Popularity in Britain – the Short Run Dynamics’, European Journal of Political Research, 14 (1986), 45–61CrossRefGoogle Scholar; and Sanders, D., Ward, H. and Marsh, David (with T. Fletcher), ‘Government Popularity and the Falklands War: A Reassessment’, British Journal of Political Science, 17 (1987), 281–313.CrossRefGoogle Scholar

¹² Norpoth, H., ‘Guns and Butter and Government Popularity in Britain’, American Political Science Review, 81 (1987), 949–59CrossRefGoogle Scholar; and Clarke, H., Stewart, M. and Zuk, G., ‘Politics, Economics, and Party Popularity in Britain, 1979–1983’, Electoral Studies, 5 (1986), 123–41.CrossRefGoogle Scholar

¹³ There is some debate about the relative advantages of using monthly versus quarterly measures of political support and economic performance. Although we collected monthly measures for these variables, the small sizes of some of the early monthly samples combined with the extreme volatility in the party support data led us to use quarterly averages. This has the effect of smoothing some of the sharper spikes in the monthly data that result variously from sampling error and exogenous shocks. This enables us better to distinguish true change from error and to focus more clearly on longer-term trends.

¹⁴ Although we focus explicitly on fluctuations in the Conservative party's share of public support, it should be obvious that Labour's share of support changes in exactly the opposite direction and by an identical amount. Therefore, the models used to explain Conservative support below apply equally to Labour support but with the signs of the coefficients reversed.

¹⁵ It should be noted, however, that although our data on party support in Britain extend from the first quarter of 1964 to the third quarter (June) of 1988, complete data for all of the independent variables used in our analyses were available only up to the end of 1986. Consequently, although our descriptions of trends in British party support extend to the middle of 1988, our analysis of the causes of those trends necessarily ends eighteen months earlier.

¹⁶ Box, G. E. P. and Jenkins, G. M., Time Series Analysis: Forecasting and Control (San Francisco: Holden-Day, 1976)Google Scholar; and Box, G. E. P. and Tiao, G. C., ‘Intervention Analysis with Applications to Economic Environmental Problems’, Journal of the American Statistical Association, 70 (1975), 70–9.CrossRefGoogle Scholar For an unusually readable discussion of the procedures involved in identifying, estimating and analysing such models see McCleary, R. and Hays, R. A., Applied Time Series Analysis for the Social Sciences (Beverly Hills, Calif.: Sage, 1980).Google Scholar

¹⁷ The data on the Conservative share of public support are not stationary in their original form (i.e. they trend upward and downward over time). The usual solution is to difference the series by subtracting the observation at time_l−1 from the observation at time_l. However, this procedure resulted in what appeared to be ‘over-differencing’ of the series, producing illogical and uninterpretable results. The problem was resolved by centring the series (i.e. removing the mean) before constructing the noise model.

¹⁸ For a recent review of the literature on mid-term effects in Britain, see Hudson, J., ‘The Relationship Between Government Popularity and Approval for the Government's Record in the United Kingdom’, British Journal of Political Science, 15 (1985), 165–86.CrossRefGoogle Scholar

¹⁹ A recent study which supports the business cycle interpretation for at least one recent British government is Sanders, , Ward, and Marsh, , ‘Government Popularity and the Falklands War’Google Scholar, which argues that the strong revival of Conservative support in the spring of 1982 was substantially due to public economic expectations created through ‘intelligent (or cynical) macroeconomic management’ by the Thatcher government.

²⁰ In estimating the multivariate model it proved necessary to revise the noise component to ensure that the residuals remained appropriately free of autocorrelation. This was done by adding a moving average (MA) term of order (5). Unlike an autoregressive process, the impact of which persists over extended periods, moving average processes reflect short-lived ‘shocks’ to the time series and can be conceived as representing the effects of politically significant events not explicitly included in the multivarate model. Although important for estimating the effects of variables included in the model, the MA term has no substantive significance of its own.

²¹ The Residual Mean Square is a standard ‘goodness-of-fit’ measure and indicates the amount of variance remaining in the data after the model has been applied. In contrast, the Ljung-Box Q statistic measures ‘badness-of-fit’, indicating whether significant autocorrelations remain in the model's residuals after the noise components and interventions are applied. Finally, the pseudo R ² statistic reported here is a measure of the success with which the intervention terms ‘explain’ the variance in the Conservative share of support after the noise model is applied. It is calculated simply as one minus the ratio of the residual sum of squares from the final model (including both noise model and intervention terms) over the residual sum of squares from the noise model only.

²² It should be noted that the pseudo R ² statistic we employ measures only the extent of variance ‘explained’ by the substantive variables included in the model. It explicitly excludes the variance explained by the noise model. In this regard it is a much more conservative measure than the traditional R ² used in Ordinary Least Squares analyses. The latter, when used in time-series analyses, can be highly inflated by autocorrelation. Indeed, if we include the variance accounted for by the noise component, our model accounts for 84 per cent of the total variance in party support – a figure that corresponds very well with regression estimates using similar data.

²³ An excellent summary of the differences between the issue priority model and the standard reward/punishment model is provided by Clarke, , Stewart, and Zuk, , ‘Politics, Economics, and Party Popularity in Britain, 1979–1983’, pp. 127–30.Google Scholar See, also, Monroe, K. R., Presidential Popularity and the Economy (New York: Praeger, 1984)Google Scholar; Peffley, M., ‘The Voter as Juror: Attributing Responsibility for Economic Conditions’, Political Behavior, 6 (1984), 175–294CrossRefGoogle Scholar; and Alt, , ‘Dealignment and the Dynamics of Partisanship in Britain’, pp. 289–329.Google Scholar

²⁴ Norpoth, , ‘Guns and Butter and Government Popularity in Britain’Google Scholar, uses monthly data and estimates that the Falklands war had a gradual temporary effect on Conservative support of 11 percentage points, about half of which effect persisted up to the May 1983 election. Clarke, , Stewart, and Zuk, . in ‘Politics, Economics, and Party Popularity’Google Scholar, also use monthly data and a different control model in estimating that the Falklands war produced an abrupt, permanent (i.e. up to the election) increase in Conservative party support of about 7 percentage points. Our study differs from these previous ones in several respects. Specifically, we use quarterly averages rather than monthly data and focus on the balance of major-party support rather than simple party support. We also use a more extensive control model than other studies and focus on a much longer time period. Although we experimented with various specifications of the Falklands effect including both the abrupt-permanent model used by Clarke and the gradual-temporary model used by Norpoth, the formulation that was most effective in accounting for variation in the balance of party support was an abrupt-temporary model in which the Falklands effect began in the second quarter of 1982 and ended in the second quarter of 1983.

²⁵ Our estimate of the Falklands effect is much higher than that of Sanders, Ward, and Marsh, in ‘Government Popularity and the Falklands War’, pp. 298–312Google Scholar, who use monthly data and ordinary least squares procedures in estimating the Falklands effect as about 3 percentage points over a period of about three months. In discounting the effects of the Falklands war they attribute the surge in Conservative fortunes in 1982 to rising personal economic expectations, the unemployment rate, the exchange rate twelve months earlier and public sector borrowing requirements at time t ₋₆. Although highly provocative, their analysis is suspect in several respects, not least because of severe multicollinearity among their independent variables and the curious specification of their control model. For further discussion of these points, see Clarke, H., Mishler, W. and Whiteley, P., ‘Re-capturing the Falklands: Models of Conservative Support, 1979–83’, British Journal of Political Science (forthcoming).Google Scholar

²⁶ Thus the impact of industrial strife increases in the first quarter after the disorder occurs by an amount equal to δ₁ω₀ In the second quarter after the occurrence, it increases again, by the smaller amount, . It continues to increase, asymptotically, in successive quarters, at an increasingly diminished rate.

²⁷ Although similarly constructed and obviously related, our measures of leadership satisfaction and party support are far from being identical. The zero-order correlation between the two is only 0.54 and that probably is inflated by autocorrelation, which the ARIMA models remove.

²⁸ Because of multicollinearity, the measures of public satisfaction with the Prime Minister and the Leader of the Opposition could not be estimated simultaneously. Thus their effects were estimated in separate but otherwise identical models.

²⁹ Among the many who rely on an issue-voting model are Crewe, , ‘Why the Conservatives Won’, pp. 282–96Google Scholar; Crewe, , ‘How to Win a Landslide Without Really Trying’, pp. 176–81Google Scholar; Särlvik, and Crewe, , Decade of Dealignment, pp. 247–80Google Scholar; and Franklin, , The Decline of Class Voting.Google Scholar

³⁰ Franklin, . The Decline of Class Voting, p. 146.Google Scholar

³¹ Crewe, , ‘What's Left for Labour’, p. 55.Google Scholar

³² Miller, , ‘There Was No Alternative’, p. 375.Google Scholar